[OSM-talk] #AttributionIsNotOptional experiment on OSM France tile servers

Wed Mar 11 13:49:57 UTC 2020

Simon,

I guess with small overlap you mean it's only about people who use osm.org
tiles, not people who use other services?
While that is true, the double whammy of both heavily using resources and
also not attributing does seem like a good subgroup to start with some
measures.

In the case of the OSM.org tiles, I suppose this is regulated by
https://wiki.osmfoundation.org/wiki/Terms_of_Use . At first glance I didn't
see anything providing people who do not respect those terms. Am I missing
something, or is this a naive approach to the problem?
Even if the ToU's could be lacking in detail, couldn't we simply change
them? The final section talks about changes, which we seem to be able to
just do when we want to.

I would think the biggest challenge on OSMF side would be the workload for
OWG/sysadmins.

Best,
Joost

Op zo 8 mrt. 2020 om 12:18 schreef Simon Poole <simon at poole.ch>:

> Just for the record:
>
> Enforcing attribution for services that you are providing directly (aka
> tiles in some form) only has a small overlap with the goals of the
> attribution guideline, and the avenues open to you depend on your ToUs /
> contracts with your users and the legal situation in the countries you are
> providing the service in.
>
> I would be very very wary of doing anything that deliberately defaces a
> web site without consulting with a local (to the country the web site is
> in) lawyer, particularly if the message implies wrong doing. The safe, I
> admit also the less fun, option, is to simply block access after giving any
> required notice.
>
> Simon
> Am 08.03.2020 um 11:04 schrieb Yves:
>
> This looks at first as a nuisance that could be perceived as a bad move,
> but the feedback you're receiving rather prove the contrary.
> Well done!
> Ps: would you share your nginx partial redirect, I may consider it for
> Opensnowmap tiles policy?
>
> Le 8 mars 2020 10:14:58 GMT+01:00, Christian Quest
> <cquest at openstreetmap.fr> <cquest at openstreetmap.fr> a écrit :
>>
>> Here is a hort report on this experiment...
>>
>> I started a week ago by searching OSM France tile server logs for referer
>> and checked manually if the map on the refering page was correctly
>> attributed.
>>
>> This allowed me to create a short list of 20 entries of sites using the
>> french styled tiles and the humanitarian tiles (yes, it is made by OSM
>> France).
>>
>>
>> I then modified our nginx based proxy_cache configuration, to redirect
>> some tiles to an "attribution tile" only for the domain in the list.
>>
>> For two of them, I tweeted about it... the most visible one is the moroco
>> yellow page service, generating a little less than a million daily tile
>> requests on our servers.
>>
>> https://twitter.com/cq94/status/1234516075695525888
>>
>> In less than 24 hours, the attribution appeared and I removed them from
>> the list.
>>
>> https://twitter.com/cq94/status/1234779931537739776
>>
>>
>> Then I included an email address in the attribution reminder tile... and
>> got emails back within a few hours.
>>
>> Some were asking how to do the attribution, others telling me the
>> attribution was now ok and asking how to remove the reminder tiles.
>>
>> In my answers, I also remind that our tile service made by volunteers on
>> donated hardware is not unlimited and inviting them to have a look at
>> switch2osm to setup their own tile server or use a commercial provider.
>>
>> Up to now, nobody complained :)
>>
>>
>> Yesterday, I've started automating attribution checking using selenium.
>> For each referer, a python script loads the page, searches for tiles, then
>> looks for attribution text or link. The result is stored in a postgresql
>> database which allows to group referers by url, hostname and ip.
>>
>> The attribution percentage I currently see is around 70-80% which is not
>> that bad.
>>
>> My next major step is to use the same technique to remind about tile
>> usage policy...
>>
>>
>> To do something similar on osm.org, a first step is to extract referers
>> from the cache logs, then use the automated attribution check to evaluate
>> the situation.
>>
>>
>> Le 08/03/2020 à 01:52, Nuno Caldeira a écrit :
>>
>> That would be a good option for those that use third party providers of
>> OSM. But to be honest, from my experience I highly doubt that even
>> corporate members of OSMF, like Mapbox would do it, when their client
>> Facebook (also corporate member of OSMF) after one year and half, still has
>> maps with lack of attribution or attributed to HERE, when it's clearly OSM.
>>
>> On Sun, 8 Mar 2020, 00:46 Phil Wyatt, <phil at wyatt-family.com> wrote:
>>
>>> I am sure others may have seen this 'blacklist' implementation for
>>> showing a reminder about attribution.
>>>
>>> https://twitter.com/cq94/status/1234528717604577282
>>>
>>> Worthy of consideration for openstreetmap.org?
>>>
>>> Cheers - Phil
>>
>> --
>> Christian Quest - OpenStreetMap France
>>
>>
> _______________________________________________
> talk mailing listtalk at openstreetmap.orghttps://lists.openstreetmap.org/listinfo/talk
>
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk
>

-- 
Joost Schouppe
OpenStreetMap <http://www.openstreetmap.org/user/joost%20schouppe/> |
Twitter <https://twitter.com/joostjakob> | LinkedIn
<https://www.linkedin.com/pub/joost-schouppe/48/939/603> | Meetup
<http://www.meetup.com/OpenStreetMap-Belgium/members/97979802/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20200311/0cc02f6d/attachment.htm>